Accumulation of mitochondrial DNA deletions is observed especially in dopaminergic neurons of the substantia nigra during ageing and even more in Parkinson’s disease. The resulting mitochondrial dysfunction is suspected to play an important role in neurodegeneration. However, the molecular mechanisms involved in the preferential generation of mitochondrial DNA deletions in dopaminergic neurons are still unknown. To study this phenomenon, we developed novel polymerase chain reaction strategies to detect distinct mitochondrial DNA deletions and monitor their accumulation patterns. Applying these approaches in in vitro and in vivo models, we show that catecholamine metabolism drives the generation and accumulation of these mitochondrial DNA mutations. As in humans, age-related accumulation of mitochondrial DNA deletions is most prominent in dopaminergic areas of mouse brain and even higher in the catecholaminergic adrenal medulla. Dopamine treatment of terminally differentiated neuroblastoma cells, as well as stimulation of dopamine turnover in mice over-expressing monoamine oxidase B both induce multiple mitochondrial DNA deletions. Our results thus identify catecholamine metabolism as the driving force behind mitochondrial DNA deletions, probably being an important factor in the ageing-associated degeneration of dopaminergic neurons.
The two major pathological hallmarks of Parkinson’s disease are the presence of intracellular protein aggregates termed Lewy bodies, and the specific loss of dopaminergic neurons in the A9 region of the substantia nigra pars compacta, which innervate the dorsal striatum. In humans, these neurons contain the black pigment neuromelanin. Their loss causes a depletion of dopamine in the striatum, leading to the well-described motor symptoms of the disease. Interestingly, vulnerability of these neurons is rather selective, as other populations of dopaminergic neurons, e.g. those of the ventral tegmental area (A10) are spared from cell death (Dauer and Przedborski, 2003).
Even though the exact mechanisms leading to Parkinson’s disease still have to be elucidated, several pan-cellular risk factors have been identified, ageing being the strongest one (Calne and Langston, 1983; de Lau and Breteler, 2006). Indeed, Parkinson’s disease is often referred to as an ageing disease, as its onset is at a mean age of 55 years and its incidence rises exponentially after 65 years (de Rijk et al., 1997; de Lau and Breteler, 2006). Other pan-cellular risk factors include exposure to environmental toxins of yet unknown nature, with 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP) or rotenone being well-described examples, and mutations in several genes, which lead to early-onset and inheritable forms of Parkinson’s disease (Schapira, 2008). As all these factors are likely to affect a broad range of cell types, research has focused on identifying the cell-specific factors that might explain the puzzling selective vulnerability of dopaminergic neurons of the substantia nigra pars compacta (Liss et al., 2005; Surmeier et al., 2012). Damaging molecules generated by dopamine metabolism, such as reactive oxygen species and oxidized intermediates of dopamine (Spencer et al., 2002; Sulzer, 2007) as well as neuromelanin, have been suspected to play an important causal role in the disease (Fasano et al., 2006; Sulzer, 2007), but conflicting reports still question their involvement (Braak et al., 2003; Zecca et al., 2008). Also, the reasons for the growing risk to develop idiopathic Parkinson’s disease with increasing age are unclear. However, there is accumulating evidence that a decline in mitochondrial function over the lifetime is involved (Schapira, 2008). Indeed, mitochondrial dysfunction is considered to be a major driving force for dopaminergic neuron loss, as MPTP or rotenone causing Parkinson’s disease symptoms impair mitochondrial respiratory chain function (Przedborski et al., 2001), and many of the mutations responsible for inheritable Parkinson’s disease are found to affect gene products (PINK1, parkin, DJ1) which are involved in mitochondrial turnover and quality control (Narendra et al., 2008; Guzman et al., 2010; Ashrafi and Schwarz, 2012; McCoy and Cookson, 2012). Finally, and importantly strengthening the link between ageing and mitochondrial dysfunction in Parkinson’s disease, two studies (Bender et al., 2006; Kraytsberg et al., 2006) have clearly shown that large deletions of mitochondrial DNA accumulate with increasing age in single dopaminergic neurons of the substantia nigra pars compacta in the human brain, and even more in patients with Parkinson’s disease (Bender et al., 2006). This continuously increasing load of mutated mitochondrial DNA copies over time may thus be an important factor contributing to dopaminergic neuron death in old age. Indeed, in both studies, deletions were reaching levels high enough to cause mitochondrial dysfunction, as confirmed at the single cell level by loss of cytochrome c oxidase activity. In addition, it was shown that most neurons contain only one type of deleted species (Reeve et al., 2008), as previously observed in single cardiomyocytes of old hearts (Khrapko et al., 1999), which suggests a clonal expansion of deleted molecules over time during ageing, probably through random drift (Elson et al., 2001; Wiesner et al., 2006).
Given the potential and causal role of mitochondrial DNA deletions in the pathogenesis of Parkinson’s disease, it appears crucial to understand why they accumulate preferentially in the substantia nigra pars compacta. However, even though this phenomenon was convincingly shown for the first time two decades ago (Cortopassi and Arnheim, 1990; Corral-Debrinski et al., 1992; Soong et al., 1992) and recently confirmed (Meissner et al., 2008), the mechanisms involved remain unclear. Therefore, we determined by complementary in vivo and in vitro approaches whether a high rate of catecholamine metabolism is responsible for the preferential generation and/or expansion of mitochondrial DNA deletions in dopaminergic neurons.
Materials and methods
Mouse tissue samples
Animals of the strain C57BL/6 were kept under standard laboratory conditions, in compliance with protocols that were approved by local government authorities (Landesamt für Natur, Umwelt- und Verbraucherschutz, LANUV, Recklinghausen) and were in accordance with NIH guidelines. Generation of mice overexpressing monoamine oxidase B in astrocytes has been previously described (Mallajosyula et al., 2008). After cervical dislocation, brain and adrenal glands were removed and immediately snap-frozen in liquid nitrogen for further processing. Adrenal glands were either used in toto or were manually dissected to produce samples of cortex and medulla.
Histological staining procedures
Frozen sections of adrenal gland (10 µm) were assessed for mitochondrial activity by evaluating cytochrome-c oxidase enzymatic activity as previously described (Sciacco et al., 1996). Tyrosine hydroxylase immunoreactivity was performed using a rabbit polyclonal primary antibody (1:750 dilution, Abcam ab112) and a rhodamine-coupled goat anti-rabbit secondary antibody (1:200 dilution, Jackson Immunoresearch, 111-025-144). Nuclei were counterstained with DAPI (Sigma Aldrich).
Cell culture and transfections
Human neuroblastoma SH-SY5Y cells (Deutsche Sammlung von Mikroorganismen und Zellkulturen, DSMZ) were grown under standard conditions in Dulbecco’s modified Eagle medium: Nutrient Mixture F-12 (DMEM/F-12), 1× non-essential amino acids, streptomycin (100 µg/ml), penicillin (100 U/ml, all Life Technologies) supplemented with uridine (200 µM, Sigma-Aldrich) and 15% heat-inactivated foetal calf serum (vol/vol, PAA). Cells were differentiated in a two-step procedure [modified from Encinas et al. (2000)], alltrans-retinoic acid (10 µM, Sigma-Aldrich) was added 1 day after seeding. After 5 days in the presence of alltrans-retinoic acid, cells were washed once and further incubated with 2 ng/ml brain-derived neurotrophic factor (Sigma-Aldrich) in Dulbecco’s modified Eagle medium/F-12 without serum for up to four additional weeks. For transfections, coding sequences of the dopamine transporter (DAT; human solute carrier Family 6 Member 3, Entrez Gene ID: 6531, SLC6A3), monoamine oxidase A (Entrez Gene ID: 4128, MAOA) and monoamine oxidase B (Entrez Gene ID: 4129, MAOB) were cloned into pcDNA4 vector (Life Technologies). Inserts were sequence-verified before transfection. SH-SY5Y cells were electroporated using Cell Line Nucleofector Kit V (Lonza). Clones were selected with 200 µg/ml zeocin (Life Technologies) and only clones with a differentiation potential similar to wild-type cells were used for further experiments.
Dopamine (20 µM, Sigma-Aldrich) was added freshly every 48 h. To avoid any auto-oxidation of dopamine, ascorbic acid (1 mM, Sigma-Aldrich) was also added.
DNA preparation, polymerase chain reaction and gel extraction
Total DNA was prepared from mouse tissues using DNeasy® Blood & Tissue Kit or QIAamp® DNA Micro Kit for smaller samples (Qiagen). Using extraction kits had no effect on mitochondrial DNA deletion levels compared with traditional phenol chloroform extraction (data not shown). DNA was extracted from SH-SY5Y cells with the High Pure PCR Template Preparation Kit (Roche). PCR and gel extraction were performed using the JumpStart™ Taq ReadyMix™ (Sigma-Aldrich) and the GeneJET gel extraction kit (Thermo Scientific), respectively. All PCR conditions are available on request.
RNA preparation and reverse transcription
Total RNA from SH-SY5Y cells was isolated with PureLink® RNA Mini Kit (Ambion) and subsequently used for reverse transcription with the QuantiTect® Reverse Transcription Kit (Qiagen). For single mouse neurons, reverse transcription was performed directly with 5 µl of cell-water suspension (see below) using the SuperScript® VILO™ cDNA Synthesis Kit (Life Technologies).
Mitochondrial DNA deletion screening strategy
For the screening of mitochondrial DNA deletions in mouse tissue samples and single neurons, primers encompassing previously reported deletion regions (Tanhauser and Laipis, 1995) were designed (PerlPrimer v1.1.18) and tested by PCR on total DNA samples from aged mouse brain. The obtained PCR products were cloned into a pJET2.1 vector (CloneJET PCR cloning Kit, Thermo Scientific) and after sequence confirmation, the three most consistently found deletions were selected for further analysis [named A, B and C, identical to deletions 1, 3 and 17 in Tanhauser and Laipis (1995)]. Primers were designed to generate specific 180-bp amplicons of these deletions as well as of the non-deleted D-Loop region (Fig. 1A and Supplementary Table 1) and subsequently used either for quantitative PCR (tissues) or for PCR (single neurons, using 5 µl of cell-water suspension, see below). Human SH-SY5Y cells were screened for deletions using a nested PCR approach to increase the sensitivity and specificity. Nested primer pairs (Supplementary Fig. 1 and Supplementary Table 1) encompassing 10 previously reported deletions (www.mitomap.org) covering the whole mitochondrial DNA were designed. For each deletion, the outer primers were first used in 30 cycles of PCR with 25 ng of total DNA as the starting template. A second 40 cycles of PCR was then performed using the inner primers and a 1:25 dilution of the first reaction as template. To allow us to accurately compare results, agarose percentage (1.5%) as well as the volume of loaded samples (5 µl for control D-Loop and 10 µl for each deletion, from 25 µl reactions) were always the same. Prominent bands were eluted and cloned for subsequent sequencing, as described above.
Quantitative polymerase chain reaction
Quantitative determination of mitochondrial DNA deletions was performed with 25 ng total DNA using QuantiTect® SYBR® Green PCR Kit (Qiagen) and the 7500 fast real-time PCR system (Applied Biosystems) with three technical replicates. PCR efficiencies were calculated with LinRegPCR software (LinRegPCR v11.1, Ramakers et al., 2003). Ratios of deleted mitochondrial DNA to wild-type mitochondrial DNA were determined with the ΔCT method, relating the non-deleted D-Loop region as indicator for total mitochondrial DNA copy number to the three individual deletion products. Product identity and integrity were confirmed on agarose gels and through melting curve analysis. ΔCT values correlated to template concentration over a range of four magnitudes with a correlation coefficient >0.96 and PCR efficiencies for the chosen primers were >99% (Supplementary Fig. 2 and Supplementary Table 2).
Single cell harvesting of dopaminergic neurons from brain slices
For single cell PCR, we harvested dopaminergic neurons from acute brain slices from female and male mice (C57BL/6). The preparation and electrophysiological recordings were performed according to Konner et al. (2011). Recordings were performed in the whole-cell configuration. The extracellular solution contained (in mM): 125 NaCl, 2.5 KCl, 2 MgCl2, 2 CaCl2, 1.2 NaH2PO4, 21 NaHCO3, 10 HEPES, and 5 glucose adjusted to pH 7.2 (with NaOH) resulting in an osmolarity of ∼310 mOsm. The intracellular solution contained (in mM): 128 K-gluconate, 10 KCl, 10 HEPES, 0.1 EGTA, 2 MgCl2, 3 K-ATP, 0.3 Na-GTP and adjusted to pH 7.3 (with KOH) resulting in an osmolarity of ∼300 mOsm. Dopaminergic neurons were ‘pre’-identified by their large cell bodies, their slow and regular firing and/or the presence of a large Ih-dependent ‘sag’-potential (Lacey et al., 1989; Richards et al., 1997; Ungless et al., 2001). In all cases their dopaminergic status was confirmed by showing the presence of Th messenger RNA. After pre-identification, the cell content was aspirated into the pipette as completely as possible. Then, the pipette was quickly removed from the brain slice and the contents were transferred to a test tube containing 10 µl RNase-free water with 250 units/ml RNase-inhibitor (Sigma-Aldrich). The test tube was immediately stored in liquid nitrogen. Cell suspensions were subsequently split for reverse transcriptase PCR and mitochondrial DNA deletion screens (see above).
Unpaired, two-tailed t-tests and Fisher’s exact test were performed using GraphPad Prism version 4.00 for Windows (GraphPad Software, www.graphpad.com). In all the figure panels, asterisks denote statistically significant changes from control, and error bars indicate standard deviations (SD) from at least three experiments, if not stated otherwise. Quantitative PCR was always performed at least in triplicate per individual sample. The P-values were calculated from unpaired t-tests. Asterisks are defined as *P < 0.05; **P < 0.01; and ***P < 0.001.
Concerning regression analysis (Supplementary Fig. 2A), when mitochondrial DNA deletions were not detectable, arbitrary abundance values were generated as reported elsewhere (Meissner et al., 2008) to allow for regression analysis.
Presence of neuromelanin is not necessary for the preferential accumulation of mitochondrial DNA deletions
To investigate the possible roles of dopamine metabolism and neuromelanin in the generation and/or expansion of mitochondrial DNA deletions, we determined the pattern of accumulation of three previously reported deletions (Tanhauser and Laipis, 1995) in the substantia nigra as well as three other brain regions from a cohort of mice between 5 and 167 weeks. This appeared to us a convenient approach to study the influence of neuromelanin, as dopaminergic neurons of mice do not contain this pigment (Marsden, 1961; Barden and Levine, 1983). The three chosen deletions were the ones we found most frequently in a preliminary screen of aged mouse brains (Fig. 1A). Quantitative analysis showed that, although all brain regions increasingly accumulate mitochondrial DNA deletions with ageing, substantia nigra and striatum, which is the projection area of the mesostriatal pathway, were the most affected (Fig. 1B). Determination of mitochondrial DNA copy number for young (5 weeks) and old (100 weeks) analysed mice samples revealed a decrease in mitochondrial DNA content with ageing for substantia nigra, striatum and cortex (Supplementary Fig. 1). However, as cerebellum had comparable levels of mitochondrial DNA copies, this suggests that a negative correlation between mitochondrial DNA deletions load and copy number is unlikely (see also the following results on adrenal gland). In the histogram of Fig. 1B, values for all three deletions have been added and related to the signal derived from the D-Loop region that is necessary for propagation of mitochondrial DNA molecules, thus giving a true percentage of molecules containing at least one of these three deletions. The highest level of deletions (0.01% of all molecules) was observed in the substantia nigra of one old mouse (167 weeks). This value is much lower than the ones previously observed in substantia nigra from very old humans, where deletion levels can reach between 0.5% and 1% of all mitochondrial DNA molecules (Corral-Debrinski et al., 1992; Soong et al., 1992; Meissner et al., 2008). Guo et al. (2010a) also found similar low deletion levels in mouse substantia nigra, and concluded that this region is much less affected than in humans. However, assuming an exponential increase, as suggested by our own data (Supplementary Fig. 2A and Supplementary Table 1), regression analysis of the data for substantia nigra showed that the amount of mitochondrial DNA deletion A doubles within 24 weeks in mice (Supplementary Table 1), which is 10 times faster than for the 4977-bp common deletion in humans (Meissner et al., 2008). Even though this faster accumulation is probably a result of the different metabolic rate between the two species, it is nevertheless remarkable that the 167-week-old mouse displayed mitochondrial DNA deletion levels in substantia nigra that are observed in 20-year-old humans (Del A: 0.006%). Interestingly, the deletion distribution pattern was similar between all brain regions analysed, with deletion A being the most prominent (Supplementary Fig. 2B).
From these first results, we can conclude that in mice, similarly to humans, brain regions with highly active dopamine metabolism preferentially accumulate mitochondrial DNA deletions, whereas the cerebellum, for example, is clearly less affected. More importantly, it means that neuromelanin, which has been discussed as a pro-oxidant (Fasano et al., 2006), can be excluded as being involved in this preferential accumulation, since it is absent in mouse nigral neurons.
Enhanced dopamine turnover in astrocytes leads to increased mitochondrial DNA deletion levels in substantia nigra
Previously, a mouse model with inducible over-expression of monoamine oxidase B preferentially present in astrocytes, was developed, displaying an increased vulnerability of dopaminergic neurons to MPTP toxicity (Mallajosyula et al., 2008). We investigated whether the enhanced turnover of dopamine, already shown to occur in these brains (Mallajosyula et al., 2008), was accompanied by an increase of mitochondrial DNA deletions levels in 50-week-old mice. Indeed, even though monoamine oxidase B over-expression elicited an increase of deleted mitochondrial DNA in all brain regions, only in substantia nigra mitochondrial DNA were deletion levels significantly increased (14-fold) compared to non-induced controls (Fig. 1C). These results suggest that both neuron intrinsic and extrinsic turnover of dopamine are a severe threat for mitochondrial DNA integrity in substantia nigra.
Mitochondrial DNA deletions preferentially accumulate in the catecholaminergic adrenal medulla
As most mitochondrial DNA deletions were found in brain regions with highly active dopamine metabolism, we postulated that a high rate of catecholamine turnover is sufficient to drive their generation and/or expansion. To test this, we chose to analyse deletions in mouse adrenal gland, a catecholaminergic tissue. Strikingly, we found that the ratio of deleted versus wild-type mitochondrial DNA was 5–10 times higher in this tissue compared with substantia nigra (Fig. 2A). Deletion levels in the adrenal gland of the 167-week-old mouse were even reaching those observed in the substantia nigra of 40-year-old humans (Del C: 0.07%; Meissner et al., 2008). Mitochondrial copy number determination in young and aged adrenal gland showed no significant change with ageing, again strengthening our assumption that there is no correlation between deletion levels and mitochondrial DNA content (Supplementary Fig. 1). Moreover, regression analysis showed that the accumulation of mitochondrial DNA deletions was even faster in the adrenal gland compared to substantia nigra and striatum (doubling time = 15 weeks; Supplementary Table 1). Remarkably, we observed a flagrant difference in the accumulation patterns between brain tissues and adrenal gland. While deletion A (3881-bp) was most prominent in all brain sections, in adrenal gland the minor arc-located deletion C (3840-bp) was prevalent (Supplementary Fig. 2B). In both cases, deletion B (3739-bp) was the least abundant species. As the synthesis of the catecholamines epinephrine and norepinephrine occurs in chromaffin cells of the adrenal medulla (Fig. 2B), we hypothesized that most mitochondrial DNA deletions are generated in this region. To test this, cortex and medulla samples were isolated and mitochondrial DNA copy number was determined for both regions. We found that, due to its high content of steroid hormone synthesizing mitochondria (Fig. 2C), cortex contains about two times more mitochondrial DNA copies than medulla (Fig. 2D). However, in these 25-week-old mice, mitochondrial DNA deletion levels were five times higher in the medulla than in the cortex (Fig. 2E).
Taken together, these results strongly support our hypothesis of a causative role of catecholamine metabolism in the accumulation of mitochondrial DNA deletions during ageing.
Dopaminergic metabolism drives the generation of mitochondrial DNA deletions in SH-SY5Y cells
To further demonstrate that enhanced catecholamine metabolism is indeed sufficient to generate mitochondrial DNA deletions, we used the human SH-SY5Y neuroblastoma cell line, which displays several characteristics of dopaminergic neurons such as the expression of tyrosine hydroxylase, the dopamine transporter (DAT) as well as the two isoforms of the dopamine degrading enzyme monoamine oxidase, monoamine oxidase A and monoamine oxidase B (Lopes et al., 2010). In addition, they can be differentiated into neuron-like structures when sequentially treated with retinoic acid and brain-derived neurotrophic factor (Encinas et al., 2000) (Supplementary Fig. 3A). We produced clones stably over-expressing three of these key proteins (DAT, monoamine oxidase A and monoamine oxidase B). Overexpression was confirmed by quantitating transgene expression by reverse transcriptase PCR (Supplementary Fig. 3B), as well as measuring monoamine oxidase enzymatic activities (data not shown). For mitochondrial DNA deletion screening, we designed a panel of 10 nested primer pairs spanning the whole human mitochondrial genome (Supplementary Fig. 1 and Supplementary Table 1). These were specific for mitochondrial DNA, as witnessed by the absence of PCR products in 143B rho zero osteosarcoma cells, which completely lack mitochondrial DNA (King and Attardi, 1989) (Supplementary Fig. 3C). Choosing this method enables one to monitor mitochondrial DNA deletion generation and accumulation in vitro.
In line with our hypothesis that dopamine is detrimental for mitochondrial DNA integrity, faint PCR products derived from mitochondrial DNA deletions were observed in the control clone (mock), indicating that these are even generated in differentiated SH-SY5Y cells kept under basal cell culture conditions for 2 weeks (Fig. 3A, right panels). However, clearly more PCR products were detected in the clones over-expressing DAT or one of the monoamine oxidases. Furthermore, when differentiated cells were treated with dopamine over 2 weeks at a non-toxic dose (25 µM; Supplementary Fig. 3D), the abundance of different individual PCR products was markedly enhanced in all four clones (Fig. 3A, left panels). We eluted distinct bands and, in addition, as we suspected the more diffuse gel lanes to include multiple deletions of different sizes, we analysed several of these lanes by sequencing all PCR products present in the reaction, and could confirm that all detected molecules were mitochondrial DNA deletions (Supplementary Fig. 3 and Supplementary Table 2). The breakpoints were mainly enclosed by short or imperfect repeats. Except in one case, we have never identified deletions spanning the origin of light strand of replication (OriL), consistent with its suggested crucial role for mitochondrial DNA maintenance (Wanrooij et al., 2012). Again, this shows that the chosen approach detects true mitochondrial DNA deletions and does not produce PCR artefacts. To strengthen the nested PCR results with quantitative data, we established time-course experiments and followed the progression of mitochondrial DNA deletion accumulation in differentiated wild-type and DAT over-expressing cells, by counting the number of PCR products generated at various time points (Fig. 3B and Supplementary Fig. 4). With this approach, we could confirm that DAT over-expression leads to a significant increase in deletion levels compared with wild-type cells (+45% at Day 2, +90% at Day 7). Similarly, dopamine treatment elicited increased deletions levels in both cell types (+30% at Day 7 for wild-type, +50% at Days 2 and 5 for DAT cells). More than 7 days of dopamine treatment resulted in blurred lanes that would not allow for an accurate counting of the PCR products.
These in vitro experiments supplement our in vivo observations and clearly show that dopamine metabolism is sufficient to drive the generation and/or expansion of mitochondrial DNA deletions.
Vulnerability of substantia nigra neurons is not caused by the preferential accumulation of mitochondrial DNA deletions
Different populations of dopaminergic neurons exist in the midbrain, with a subset being highly vulnerable and most prone to degeneration during ageing and Parkinsons’s disease (Uhl et al., 1985; Damier et al., 1999; Braak et al., 2003; Surmeier, 2007). In particular, as neurons in the ventral tegmental area region are more or less spared, we tested the hypothesis that they are less affected by mitochondrial DNA deletions during ageing. Different populations of neurons were identified according to their location in the midbrain and their electrophysiological properties (see ‘Materials and methods’ section). The dopaminergic status was confirmed by showing the presence of tyrosine hydroxylase messenger RNA (Fig. 4B). The presence of deletions A, B and C was investigated in 82 tyrosine hydroxylase-positive neurons by conventional PCR (Fig. 4A). None of the 24 cells from young mice (5 weeks) showed deletions, whereas 59% (34/58) of the neurons from old mice (>2 years) contained at least one of the three investigated deletions (Fig. 4C). Comparing the proportion of neurons containing deletions in the ventral tegmental area versus substantia nigra from the same old mice, we found that both regions were similar (67% in ventral tegmental area, 53% in substantia nigra).
As dopaminergic neurons from ventral tegmental area and substantia nigra pars compacta are thus equally affected by mitochondrial DNA deletions, our results imply that these alterations alone cannot account for the selective vulnerability of nigral neurons. Thus, mitochondrial DNA deletions are more likely to critically affect these neurons when combined with other cell-specific parameters.
In this study, we provide strong experimental evidence for a causal role of dopamine metabolism in the accumulation of mitochondrial DNA deletions observed in dopaminergic neurons during ageing and Parkinson’s disease. First, we show that mitochondrial DNA deletions preferentially accumulate in dopaminergic brain areas in mice, similar to humans, and that another catecholaminergic organ, namely the adrenal medulla, is also affected by high deletion levels. Second, an enhanced dopamine turnover in astrocytes leads to higher mitochondrial DNA deletion loads in substantia nigra. Last, in vitro data clearly demonstrate the deleterious influence of dopamine on mitochondrial DNA integrity. Indeed, direct exposure of cells to dopamine or over-expression of key enzymes of dopaminergic metabolism led to increased levels of mitochondrial DNA deletions. In addition, we exclude neuromelanin as being the main driving force behind mitochondrial DNA deletion accumulation, since mouse nigral neurons are devoid of the black pigment.
Dopaminergic metabolism has long been considered as one of the main causes for the observed loss of dopaminergic neurons in Parkinson’s disease. It is known that dopamine readily oxidizes and leads to the generation of reactive oxygen species, such as H2O2, OH• and O•−2 (Halliwell and Gutteridge, 1984), as well as neurotoxic quinones. Dopamine quinones can interact with cysteinyl residues of proteins (Fornstedt et al., 1986), whereas hydroxyl radicals can modify all kinds of cellular components. Indeed, oxidatively modified nuclear DNA has been observed in neurons challenged with dopamine in culture (Spencer et al., 2002), whereas cysteinyl-dopamine compounds have been detected in human dopaminergic brain areas (Rosengren et al., 1985). Here we show that dopamine is also damaging mitochondrial DNA, as witnessed by the accumulation of a wide range of deletions. One likely hypothesis is a direct deleterious effect of the neurotransmitter and/or its metabolites on mitochondrial DNA. Dopamine, which induces DNA strand breaks in vitro (Moldeus et al., 1983; Li and Cao, 2002), has been shown to access the mitochondrial matrix (Gautam and Zeevalk, 2011), whereas H2O2, produced locally by monoamine oxidases at the outer mitochondrial membrane could diffuse into the mitochondria and cause mitochondrial DNA single and double strand breaks, as demonstrated previously in vitro (Shokolenko et al., 2009). Alternatively, because of the vicinity of these reactive molecules to proteins of the outer membrane involved in mitochondrial dynamics and mitophagy, such as mitofusins, MIRO, PINK1 or parkin, respectively (Ashrafi and Schwarz, 2012), these could also be altered, thus interfering with quality control mechanisms aimed at eliminating damaged mitochondria. Interestingly, these mechanisms involving fusion/fission and autophagy are indeed importantly involved in the clearance of deleted mitochondrial DNA molecules, as shown by the accumulation of deletions in cases where mitochondrial fusion is impaired (Chen et al., 2010). Dopamine could therefore promote both the generation as well as the accumulation over time of mitochondrial DNA deletions in dopaminergic neurons.
Our strategy of monitoring three different deletions in parallel allows us to address several important questions in the field of mitochondrial genetics. In particular, until now, it was unclear why and how mitochondrial DNA deletions accumulate in cells. Also puzzling is the fact that the distribution of deletions can differ between different tissues (Guo et al., 2010b). In line with this observation, our method enabled us to find different accumulation patterns between brain and adrenal gland, suggesting that the observed accumulation of mitochondrial DNA deletions cannot be caused by random drift or a replicative advantage alone. Among alternative potential mechanisms, it has been proposed that double strand breaks coupled with irregular repair involving exonuclease activity could play a causal role in the accumulation of mitochondrial DNA deletions (Krishnan et al., 2008). In support of this, targeting the restriction enzyme PstI to mouse mitochondria induces the formation of mitochondrial DNA deletions after cutting of the double-stranded mitochondrial DNA. This enzyme cuts twice in mouse mitochondrial DNA, but deletion breakpoints different from the PstI sites were observed, indicating that an exonuclease activity has processed mitochondrial DNA starting from the restriction sites (Srivastava and Moraes, 2005). As dopamine treatment elicited not only enhanced levels of mitochondrial DNA deletions, but also an increase in the variety of species in SH-SY5Y cells, this suggests that such a mechanism could take place in dopaminergic neurons.
We also demonstrate that neuromelanin is not necessary for the preferential accumulation of mitochondrial DNA deletions in substantia nigra. Neuromelanin has been suspected to play a role in Parkinson’s disease, as mainly the dark-pigmented dopaminergic neurons of the substantia nigra pars compacta undergo neuronal death, whereas other dopaminergic regions, like the ventral tegmental area, are mainly spared. Nevertheless, its involvement in the disease is still a matter of debate, as both neurotoxic (Fasano et al., 2006) and neuroprotective (Zecca et al., 2008) properties have been reported. Contrasting with our data, neuromelanin has recently been proposed to determine the differential vulnerability of catecholaminergic neurons to mitochondrial DNA deletions, since the highest levels of deletions were found in pigmented neurons of the substantia nigra pars compacta and, to a lesser extent, of the ventral tegmental area (Elstner et al., 2011). However, this is not true for locus coeruleus, for which pigmented and non-pigmented neurons harbour similar levels of mitochondrial DNA deletions, but levels are four times lower than those observed for pigmented substantia nigra pars compacta neurons. This clearly shows that neuromelanin per se cannot be considered as the main reason behind mitochondrial DNA deletion accumulation, and suggests that cell-specific parameters are influencing this process.
Our observation that the proportions of substantia nigra and ventral tegmental area cells harbouring mitochondrial DNA deletions are similar, highlights the complexity of the selective neurodegenerative process occurring in Parkinson’s disease. Even though both types of cells are dopaminergic, and both accumulate mitochondrial DNA deletions, a subset of these neurons is much more prone to degeneration. Thus, one could argue that mitochondrial DNA deletions are not relevant for Parkinson’s disease at all. However, as they have been shown to reach critical levels severely compromising cell survival in patients with Parkinson’s disease (Bender et al., 2006), it is very unlikely that they are not involved in the pathogenesis of the disease. But, how are mitochondrial DNA deletions integrated in the chain of events leading to neuronal loss? We postulate that during life, dopaminergic neurons are constantly exposed to high levels of dopamine and its metabolites. This chronic exposure could lead to a continuous de novo generation of mitochondrial DNA deletions, as suggested by the increase in deletion species over time in SH-SY5Y cells. Nevertheless, one cannot exclude an accelerated mitochondrial DNA turnover resulting from continuous damage, which could favour the expansion of pre-existing deletions, as recently reported (Payne et al., 2011). Additionally, an ageing-related or dopamine-induced decline in mitochondrial quality control mechanisms could lead to an impaired clearance of deleted molecules. This imbalanced situation will cause an accumulation of mitochondrial DNA deletions and, when they reach a critical threshold, mitochondrial dysfunction, ultimately contributing to cell death. The differential vulnerability of nigral neurons strongly suggests that this threshold may be reached more rapidly in these cells than in ventral tegmental area neurons. One explanation for this is that substantia nigra neurons accumulate more mitochondrial DNA deletions because of higher intracellular dopamine concentrations. This hypothesis would be in line with data from Mosharov et al., (2009), who showed that cytoplasmic dopamine concentration correlates with intracellular Ca2+, which may be higher in substantia nigra neurons compared with ventral tegmental area neurons because of the sustained Ca2+ entry into the cytoplasm caused by Cav1.3 channels-driven pacemaking. Also, the proportion of neurons expressing calbindin, a Ca2+ buffering protein, is much lower in substantia nigra than in the ventral tegmental area (Liang et al., 1996; McRitchie et al., 1996), which might contribute to overall high cytosolic steady-state Ca2+ concentrations in these cells. In addition, it is conceivable that, with mitochondrial dysfunction induced by higher mitochondrial DNA deletions levels in substantia nigra pars compacta neurons, the Ca2+ buffering capacity of the organelles might as well be impaired, thus worsening the situation. Alternatively, levels of mitochondrial DNA deletions and ensuing mitochondrial dysfunction may be similar in ventral tegmental area and substantia nigra pars compacta neurons, however, the high energy requirement of Ca2+ homeostasis in the latter, and the resulting reliance on mitochondria, will make them more vulnerable (Surmeier et al., 2012).
In conclusion, we present experimental evidence for a direct role of dopamine in the generation and/or expansion of mitochondrial DNA deletions in dopaminergic neurons, thus shedding new light on the pathogenesis of Parkinson’s disease. Accumulation of mitochondrial DNA deletions might indeed be the long searched for missing link between ageing and Parkinson’s disease.
We thank Maria Bust, Brigitte Dengler, Katrin Lanz and Alexander Müller for excellent technical help.
Rudolf J. Wiesner and Peter Kloppenburg were funded by grants from the Deutsche Forschungsgemeinschaft (Cologne Excellence Cluster on Cellular Stress Responses in Ageing-associated Diseases, CECAD) and the Center of Molecular Medicine Cologne (CMMC).
Supplementary material is available at Brain online.